Memory T cells, which form after primary infection, play an important role in achieving long-term adaptive immune protection and immunosurveillance against recurring infections. Different memory CD8+ T cell populations are thought to contribute to cellular immunity at different anatomical locations: central memory (TCM) and effector memory T cells (TEM) patrol lymphoid organs and blood to provide global surveillance of pathogens, while resident memory T cells (TRM) predominantly reside in non-lymphoid tissues (NLT), which are common sites of pathogen re-encounter [1]. In both mice and humans, TRM are characterized by a high expression of the surface markers CD103 and CD69, which is known to downregulate the lymph-homing receptor S1PR1 and low expression of CD62L [2,3,4,5]. Phenotypically, TRM are thought to be similar to recently activated effector T cells as they have increased cytotoxic potential and activate local immunity through cytokine production in contrast to TCM which are re-activated in the secondary lymphoid organs [1]. This has led many to believe that TRM represent a terminally differentiated population of memory T cells permanently anchored in the tissue with little to no potential for recirculation. Over the last decade, it has become increasingly clear that T cells with a TRM-like phenotype are not entirely restricted to NTLs, but can also be found in secondary lymph organs [6,7,8].

Classically, primary and memory T cell responses are thought to be initiated in the lymph nodes from where T cells migrate into peripheral tissue sites. This model is referred to as an ‘inside-out’ model and has more recently been challenged by multiple studies [7, 9, 10]. While this model holds true for the initiation of a primary immune response, it does not align with the behavior seen in TRM. In contrast to other memory T cell populations, TRM have high proliferative potential leading to the local expansion of memory cells upon antigen restimulation [9]. Moreover, TRM may rather follow an ‘outside-in’ model of retrograde migration. This suggest that TRM can initiate a secondary immune response within tissues and then migrate to the draining lymph nodes (dLN). The presence of LN T cells with a TRM phenotype was first observed upon LCMV infection, by which parabiosis experiments indicated their tissue residency [6]. LN TRM show similar transcriptional profiles to NLT TRM and share common surface marker expression with NLT TRM. The formation of LN TRM depends on NLT TRM, which function as precursors to LN TRM, further supporting an ‘outside-in’ model of adaptive memory. However, how and when this migration from the tissue to the LN is initiated and how they re-migrate during a secondary infection has been unclear so far.

The preprint by Heim et al. [11] utilized a murine scarification model with vaccinia virus (VV) for a highly localized, self-resolving skin infection to study the ontogeny and developmental cues needed to seed resident memory CD8+ T cell populations in the skin and the dLN (Fig. 1). In line with previous reports, LN TRM form exclusively in the dLN upon VV infection and exhibit high phenotypic similarity to skin-resident counterparts such as the expression of traditional skin TRM markers CD69, CD103, and CXCR6. Moreover, skin TRM and LN TRM develop concurrently and remain randomly-distributed in the dLN up to 100 days after infection. Since skin-resident TRM may be precursors for LN TRM, the authors utilized a mouse model expressing a soluble K14-driven VEGFR3-Ig fusion protein, which leads to the loss of dermal lymphatic vessels by trapping free VEGF-C, to address whether the lymphatics are required for seeding LN TRM. While there were no changes in skin TRM abundance after VV infection, there was almost a complete ablation of LN TRM. These data suggest that dermal lymphatics are essential for the formation and seeding of LN TRM in the dLN. Using Kaede transgenic mice, which express a photoconvertible fluorescent protein, the authors tracked skin-emigrating TRM and found that LN TRM seeding primarily occurs within the first 5 days after VV infection of the ear, further supporting the notion that egressing, effector CD8+ T cells represent a potential precursor population that seeds LN TRM.

Fig. 1: CD8+ T cells seed skin and lymph node TRM after infection via the dermal lymphatic vessels.
figure 1

After infection with VV, CD8+ T cells in the skin recoginze viral antigens and subsequently seed skin TRM and LN TRM. During the effector phase, lymphatic-dependent egress of LN TRM is essnetial for LN TRM seeding. Moreover, LN TRM has a protective role upon rechallenge. Created with BioRender.

To determine whether transit through the skin is necessary for LN TRM development, the authors first depleted circulating T cells by injecting their mice with an anti-CD90.1 antibody, leaving skin TRM intact. This was then followed by surgically removing the infected ear at different timepoints to physically prevent migration from the skin to the LN. In this setup, few LN TRM were formed if the ear was resected at day 5 or later compared to non-resected controls. These data suggest that LN TRM exclusively migrate from skin to the dLN. This was further corroborated in subsequent experiments where both ears from the same mouse were infected with VV and only the non-resected infected ear was able to form LN TRM. Together, this supports the notion that T cell egress from the skin is necessary for LN TRM formation. In addition, adoptive transfer of TCR-reporter T cells (Nur77-GFP) revealed that TCR triggering occurs almost exclusively in the skin and not in the dLN, even after re-challenge. Lastly, the authors demonstrated that LN TRM are primed for cytotoxicity and support the control subsequent infections. Collectively, this study highlights that TRM transit from the skin to the dLN in the effector phase and derive from effector T cells rather than established skin TRM.

Overall, this preprint fills a gap in the mechanism of retrograde migration after skin infection by showing that dermal lymphatic vessels are required for LN TRM seeding. However, the exact environmental clues mediating which TRM leaves the skin to seed the LN are still unknown and warrant further investigation. Additionally, it shows how antigen encounter does not happen in the dLN for these TRM but rather in the skin, providing additional evidence for an ‘outside-in’ model of TRM -mediated immune memory. This study adds to the dogma that not only effector and central memory can be assessed by looking at secondary lymphoid tissues but that these organs do in fact, at least partly, represent the peripheral immune memory in NLT.